archival/unzip.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini unzip implementation for busybox
   4  *
   5  * Copyright (C) 2004 by Ed Clark
   6  *
   7  * Loosely based on original busybox unzip applet by Laurence Anderson.
   8  * All options and features should work in this version.
   9  *
  10  * Licensed under GPLv2 or later, see file LICENSE in this source tree.
  11  */
  12 /* For reference see
  13  * http://www.pkware.com/company/standards/appnote/
  14  * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
  15  *
  16  * TODO
  17  * Zip64 + other methods
  18  */
  19 //config:config UNZIP
  20 //config:       bool "unzip (24 kb)"
  21 //config:       default y
  22 //config:       help
  23 //config:       unzip will list or extract files from a ZIP archive,
  24 //config:       commonly found on DOS/WIN systems. The default behavior
  25 //config:       (with no options) is to extract the archive into the
  26 //config:       current directory.
  27 //config:
  28 //config:config FEATURE_UNZIP_CDF
  29 //config:       bool "Read and use Central Directory data"
  30 //config:       default y
  31 //config:       depends on UNZIP
  32 //config:       help
  33 //config:       If you know that you only need to deal with simple
  34 //config:       ZIP files without deleted/updated files, SFX archives etc,
  35 //config:       you can reduce code size by unselecting this option.
  36 //config:       To support less trivial ZIPs, say Y.
  37 //config:
  38 //config:config FEATURE_UNZIP_BZIP2
  39 //config:       bool "Support compression method 12 (bzip2)"
  40 //config:       default y
  41 //config:       depends on FEATURE_UNZIP_CDF && DESKTOP
  42 // FEATURE_UNZIP_CDF is needed, otherwise we can't find start of next file
  43 // DESKTOP is needed to get back uncompressed length
  44 //config:
  45 //config:config FEATURE_UNZIP_LZMA
  46 //config:       bool "Support compression method 14 (lzma)"
  47 //config:       default y
  48 //config:       depends on FEATURE_UNZIP_CDF && DESKTOP
  49 //config:
  50 //config:config FEATURE_UNZIP_XZ
  51 //config:       bool "Support compression method 95 (xz)"
  52 //config:       default y
  53 //config:       depends on FEATURE_UNZIP_CDF && DESKTOP
  54
  55 //applet:IF_UNZIP(APPLET(unzip, BB_DIR_USR_BIN, BB_SUID_DROP))
  56 //kbuild:lib-$(CONFIG_UNZIP) += unzip.o
  57
  58 //usage:#define unzip_trivial_usage
  59 //usage:       "[-lnopq] FILE[.zip] [FILE]... [-x FILE...] [-d DIR]"
  60 //usage:#define unzip_full_usage "\n\n"
  61 //usage:       "Extract FILEs from ZIP archive\n"
  62 //usage:     "\n        -l      List contents (with -q for short form)"
  63 //usage:     "\n        -n      Never overwrite files (default: ask)"
  64 //usage:     "\n        -o      Overwrite"
  65 //usage:     "\n        -j      Do not restore paths"
  66 //usage:     "\n        -p      Print to stdout"
  67 //usage:     "\n        -q      Quiet"
  68 //usage:     "\n        -x FILE Exclude FILEs"
  69 //usage:     "\n        -d DIR  Extract into DIR"
  70
  71 #include "libbb.h"
  72 #include "bb_archive.h"
  73
  74 #if 0
  75 # define dbg(...) bb_error_msg(__VA_ARGS__)
  76 #else
  77 # define dbg(...) ((void)0)
  78 #endif
  79
  80 enum {
  81 #if BB_BIG_ENDIAN
  82         ZIP_FILEHEADER_MAGIC = 0x504b0304,
  83         ZIP_CDF_MAGIC        = 0x504b0102, /* CDF item */
  84         ZIP_CDE_MAGIC        = 0x504b0506, /* End of CDF */
  85         ZIP_DD_MAGIC         = 0x504b0708,
  86 #else
  87         ZIP_FILEHEADER_MAGIC = 0x04034b50,
  88         ZIP_CDF_MAGIC        = 0x02014b50,
  89         ZIP_CDE_MAGIC        = 0x06054b50,
  90         ZIP_DD_MAGIC         = 0x08074b50,
  91 #endif
  92 };
  93
  94 #define ZIP_HEADER_LEN 26
  95
  96 typedef union {
  97         uint8_t raw[ZIP_HEADER_LEN];
  98         struct {
  99                 uint16_t version;               /* 0-1 */
 100                 uint16_t zip_flags;             /* 2-3 */
 101                 uint16_t method;                /* 4-5 */
 102                 uint16_t modtime;               /* 6-7 */
 103                 uint16_t moddate;               /* 8-9 */
 104                 uint32_t crc32 PACKED;          /* 10-13 */
 105                 uint32_t cmpsize PACKED;        /* 14-17 */
 106                 uint32_t ucmpsize PACKED;       /* 18-21 */
 107                 uint16_t filename_len;          /* 22-23 */
 108                 uint16_t extra_len;             /* 24-25 */
 109                 /* filename follows (not NUL terminated) */
 110                 /* extra field follows */
 111                 /* data follows */
 112         } fmt PACKED;
 113 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
 114
 115 #define FIX_ENDIANNESS_ZIP(zip) \
 116 do { if (BB_BIG_ENDIAN) { \
 117         (zip).fmt.method        = SWAP_LE16((zip).fmt.method      ); \
 118         (zip).fmt.crc32         = SWAP_LE32((zip).fmt.crc32       ); \
 119         (zip).fmt.cmpsize       = SWAP_LE32((zip).fmt.cmpsize     ); \
 120         (zip).fmt.ucmpsize      = SWAP_LE32((zip).fmt.ucmpsize    ); \
 121         (zip).fmt.filename_len  = SWAP_LE16((zip).fmt.filename_len); \
 122         (zip).fmt.extra_len     = SWAP_LE16((zip).fmt.extra_len   ); \
 123 }} while (0)
 124
 125 #define CDF_HEADER_LEN 42
 126
 127 typedef union {
 128         uint8_t raw[CDF_HEADER_LEN];
 129         struct {
 130                 /* uint32_t signature; 50 4b 01 02 */
 131                 uint16_t version_made_by;       /* 0-1 */
 132                 uint16_t version_needed;        /* 2-3 */
 133                 uint16_t cdf_flags;             /* 4-5 */
 134                 uint16_t method;                /* 6-7 */
 135                 uint16_t modtime;               /* 8-9 */
 136                 uint16_t moddate;               /* 10-11 */
 137                 uint32_t crc32;                 /* 12-15 */
 138                 uint32_t cmpsize;               /* 16-19 */
 139                 uint32_t ucmpsize;              /* 20-23 */
 140                 uint16_t filename_len;          /* 24-25 */
 141                 uint16_t extra_len;             /* 26-27 */
 142                 uint16_t file_comment_length;   /* 28-29 */
 143                 uint16_t disk_number_start;     /* 30-31 */
 144                 uint16_t internal_attributes;   /* 32-33 */
 145                 uint32_t external_attributes PACKED; /* 34-37 */
 146                 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 147                 /* filename follows (not NUL terminated) */
 148                 /* extra field follows */
 149                 /* file comment follows */
 150         } fmt PACKED;
 151 } cdf_header_t;
 152
 153 #define FIX_ENDIANNESS_CDF(cdf) \
 154 do { if (BB_BIG_ENDIAN) { \
 155         (cdf).fmt.version_made_by = SWAP_LE16((cdf).fmt.version_made_by); \
 156         (cdf).fmt.version_needed = SWAP_LE16((cdf).fmt.version_needed); \
 157         (cdf).fmt.method        = SWAP_LE16((cdf).fmt.method      ); \
 158         (cdf).fmt.modtime       = SWAP_LE16((cdf).fmt.modtime     ); \
 159         (cdf).fmt.moddate       = SWAP_LE16((cdf).fmt.moddate     ); \
 160         (cdf).fmt.crc32         = SWAP_LE32((cdf).fmt.crc32       ); \
 161         (cdf).fmt.cmpsize       = SWAP_LE32((cdf).fmt.cmpsize     ); \
 162         (cdf).fmt.ucmpsize      = SWAP_LE32((cdf).fmt.ucmpsize    ); \
 163         (cdf).fmt.filename_len  = SWAP_LE16((cdf).fmt.filename_len); \
 164         (cdf).fmt.extra_len     = SWAP_LE16((cdf).fmt.extra_len   ); \
 165         (cdf).fmt.file_comment_length = SWAP_LE16((cdf).fmt.file_comment_length); \
 166         (cdf).fmt.external_attributes = SWAP_LE32((cdf).fmt.external_attributes); \
 167 }} while (0)
 168
 169 #define CDE_LEN 16
 170
 171 typedef union {
 172         uint8_t raw[CDE_LEN];
 173         struct {
 174                 /* uint32_t signature; 50 4b 05 06 */
 175                 uint16_t this_disk_no;
 176                 uint16_t disk_with_cdf_no;
 177                 uint16_t cdf_entries_on_this_disk;
 178                 uint16_t cdf_entries_total;
 179                 uint32_t cdf_size;
 180                 uint32_t cdf_offset;
 181                 /* uint16_t archive_comment_length; */
 182                 /* archive comment follows */
 183         } fmt PACKED;
 184 } cde_t;
 185
 186 #define FIX_ENDIANNESS_CDE(cde) \
 187 do { if (BB_BIG_ENDIAN) { \
 188         (cde).fmt.cdf_offset = SWAP_LE32((cde).fmt.cdf_offset); \
 189 }} while (0)
 190
 191 struct BUG {
 192         /* Check the offset of the last element, not the length.  This leniency
 193          * allows for poor packing, whereby the overall struct may be too long,
 194          * even though the elements are all in the right place.
 195          */
 196         char BUG_zip_header_must_be_26_bytes[
 197                 offsetof(zip_header_t, fmt.extra_len) + 2
 198                         == ZIP_HEADER_LEN ? 1 : -1];
 199         char BUG_cdf_header_must_be_42_bytes[
 200                 offsetof(cdf_header_t, fmt.relative_offset_of_local_header) + 4
 201                         == CDF_HEADER_LEN ? 1 : -1];
 202         char BUG_cde_must_be_16_bytes[
 203                 sizeof(cde_t) == CDE_LEN ? 1 : -1];
 204 };
 205
 206
 207 enum { zip_fd = 3 };
 208
 209
 210 /* This value means that we failed to find CDF */
 211 #define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
 212
 213 #if !ENABLE_FEATURE_UNZIP_CDF
 214
 215 # define find_cdf_offset() BAD_CDF_OFFSET
 216
 217 #else
 218 /* Seen in the wild:
 219  * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
 220  * where CDE was nearly 48 kbytes before EOF.
 221  * (Surprisingly, it also apparently has *another* CDE structure
 222  * closer to the end, with bogus cdf_offset).
 223  * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
 224  */
 225 #define PEEK_FROM_END (64*1024)
 226 /* NB: does not preserve file position! */
 227 static uint32_t find_cdf_offset(void)
 228 {
 229         cde_t cde;
 230         unsigned char *buf;
 231         unsigned char *p;
 232         off_t end;
 233         uint32_t found;
 234
 235         end = lseek(zip_fd, 0, SEEK_END);
 236         if (end == (off_t) -1)
 237                 return BAD_CDF_OFFSET;
 238
 239         end -= PEEK_FROM_END;
 240         if (end < 0)
 241                 end = 0;
 242
 243         dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
 244         xlseek(zip_fd, end, SEEK_SET);
 245         buf = xzalloc(PEEK_FROM_END);
 246         full_read(zip_fd, buf, PEEK_FROM_END);
 247
 248         found = BAD_CDF_OFFSET;
 249         p = buf;
 250         while (p <= buf + PEEK_FROM_END - CDE_LEN - 4) {
 251                 if (*p != 'P') {
 252                         p++;
 253                         continue;
 254                 }
 255                 if (*++p != 'K')
 256                         continue;
 257                 if (*++p != 5)
 258                         continue;
 259                 if (*++p != 6)
 260                         continue;
 261                 /* we found CDE! */
 262                 memcpy(cde.raw, p + 1, CDE_LEN);
 263                 FIX_ENDIANNESS_CDE(cde);
 264                 /*
 265                  * I've seen .ZIP files with seemingly valid CDEs
 266                  * where cdf_offset points past EOF - ??
 267                  * This check ignores such CDEs:
 268                  */
 269                 if (cde.fmt.cdf_offset < end + (p - buf)) {
 270                         found = cde.fmt.cdf_offset;
 271                         dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
 272                                 (unsigned)found, end + (p-3 - buf));
 273                         dbg("  cdf_offset+cdf_size:0x%x",
 274                                 (unsigned)(found + SWAP_LE32(cde.fmt.cdf_size)));
 275                         /*
 276                          * We do not "break" here because only the last CDE is valid.
 277                          * I've seen a .zip archive which contained a .zip file,
 278                          * uncompressed, and taking the first CDE was using
 279                          * the CDE inside that file!
 280                          */
 281                 }
 282         }
 283         free(buf);
 284         dbg("Found cdf_offset:0x%x", (unsigned)found);
 285         return found;
 286 };
 287
 288 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf)
 289 {
 290         uint32_t magic;
 291
 292         if (cdf_offset == BAD_CDF_OFFSET)
 293                 return cdf_offset;
 294
 295         dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 296         xlseek(zip_fd, cdf_offset, SEEK_SET);
 297         xread(zip_fd, &magic, 4);
 298         /* Central Directory End? Assume CDF has ended.
 299          * (more correct method is to use cde.cdf_entries_total counter)
 300          */
 301         if (magic == ZIP_CDE_MAGIC) {
 302                 dbg("got ZIP_CDE_MAGIC");
 303                 return 0; /* EOF */
 304         }
 305         xread(zip_fd, cdf->raw, CDF_HEADER_LEN);
 306
 307         FIX_ENDIANNESS_CDF(*cdf);
 308         dbg("  filename_len:%u extra_len:%u file_comment_length:%u",
 309                 (unsigned)cdf->fmt.filename_len,
 310                 (unsigned)cdf->fmt.extra_len,
 311                 (unsigned)cdf->fmt.file_comment_length
 312         );
 313         cdf_offset += 4 + CDF_HEADER_LEN
 314                 + cdf->fmt.filename_len
 315                 + cdf->fmt.extra_len
 316                 + cdf->fmt.file_comment_length;
 317
 318         return cdf_offset;
 319 };
 320 #endif
 321
 322 static void die_if_bad_fnamesize(unsigned sz)
 323 {
 324         if (sz > 0xfff) /* more than 4k?! no funny business please */
 325                 bb_error_msg_and_die("bad archive");
 326 }
 327
 328 static void unzip_skip(off_t skip)
 329 {
 330         if (skip != 0)
 331                 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
 332                         bb_copyfd_exact_size(zip_fd, -1, skip);
 333 }
 334
 335 static void unzip_create_leading_dirs(const char *fn)
 336 {
 337         /* Create all leading directories */
 338         char *name = xstrdup(fn);
 339         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
 340                 xfunc_die(); /* bb_make_directory is noisy */
 341         }
 342         free(name);
 343 }
 344
 345 #if ENABLE_FEATURE_UNZIP_CDF
 346 static void unzip_extract_symlink(zip_header_t *zip, const char *dst_fn)
 347 {
 348         char *target;
 349
 350         die_if_bad_fnamesize(zip->fmt.ucmpsize);
 351
 352         if (zip->fmt.method == 0) {
 353                 /* Method 0 - stored (not compressed) */
 354                 target = xzalloc(zip->fmt.ucmpsize + 1);
 355                 xread(zip_fd, target, zip->fmt.ucmpsize);
 356         } else {
 357 #if 1
 358                 bb_error_msg_and_die("compressed symlink is not supported");
 359 #else
 360                 transformer_state_t xstate;
 361                 init_transformer_state(&xstate);
 362                 xstate.mem_output_size_max = zip->fmt.ucmpsize;
 363                 /* ...unpack... */
 364                 if (!xstate.mem_output_buf)
 365                         WTF();
 366                 target = xstate.mem_output_buf;
 367                 target = xrealloc(target, xstate.mem_output_size + 1);
 368                 target[xstate.mem_output_size] = '\0';
 369 #endif
 370         }
 371         if (!unsafe_symlink_target(target)) {
 372 //TODO: libbb candidate
 373                 if (symlink(target, dst_fn)) {
 374                         /* shared message */
 375                         bb_perror_msg_and_die("can't create %slink '%s' to '%s'",
 376                                 "sym", dst_fn, target
 377                         );
 378                 }
 379         }
 380         free(target);
 381 }
 382 #endif
 383
 384 static void unzip_extract(zip_header_t *zip, int dst_fd)
 385 {
 386         transformer_state_t xstate;
 387
 388         if (zip->fmt.method == 0) {
 389                 /* Method 0 - stored (not compressed) */
 390                 off_t size = zip->fmt.ucmpsize;
 391                 if (size)
 392                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 393                 return;
 394         }
 395
 396         init_transformer_state(&xstate);
 397         xstate.bytes_in = zip->fmt.cmpsize;
 398         xstate.src_fd = zip_fd;
 399         xstate.dst_fd = dst_fd;
 400         if (zip->fmt.method == 8) {
 401                 /* Method 8 - inflate */
 402                 if (inflate_unzip(&xstate) < 0)
 403                         bb_error_msg_and_die("inflate error");
 404                 /* Validate decompression - crc */
 405                 if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 406                         bb_error_msg_and_die("crc error");
 407                 }
 408         }
 409 #if ENABLE_FEATURE_UNZIP_BZIP2
 410         else if (zip->fmt.method == 12) {
 411                 /* Tested. Unpacker reads too much, but we use CDF
 412                  * and will seek to the correct beginning of next file.
 413                  */
 414                 xstate.bytes_out = unpack_bz2_stream(&xstate);
 415                 if (xstate.bytes_out < 0)
 416                         bb_error_msg_and_die("inflate error");
 417         }
 418 #endif
 419 #if ENABLE_FEATURE_UNZIP_LZMA
 420         else if (zip->fmt.method == 14) {
 421                 /* Not tested yet */
 422                 xstate.bytes_out = unpack_lzma_stream(&xstate);
 423                 if (xstate.bytes_out < 0)
 424                         bb_error_msg_and_die("inflate error");
 425         }
 426 #endif
 427 #if ENABLE_FEATURE_UNZIP_XZ
 428         else if (zip->fmt.method == 95) {
 429                 /* Not tested yet */
 430                 xstate.bytes_out = unpack_xz_stream(&xstate);
 431                 if (xstate.bytes_out < 0)
 432                         bb_error_msg_and_die("inflate error");
 433         }
 434 #endif
 435         else {
 436                 bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
 437         }
 438
 439         /* Validate decompression - size */
 440         if (zip->fmt.ucmpsize != xstate.bytes_out) {
 441                 /* Don't die. Who knows, maybe len calculation
 442                  * was botched somewhere. After all, crc matched! */
 443                 bb_error_msg("bad length");
 444         }
 445 }
 446
 447 static void my_fgets80(char *buf80)
 448 {
 449         fflush_all();
 450         if (!fgets(buf80, 80, stdin)) {
 451                 bb_perror_msg_and_die("can't read standard input");
 452         }
 453 }
 454
 455 static int get_lstat_mode(const char *dst_fn)
 456 {
 457         struct stat stat_buf;
 458         if (lstat(dst_fn, &stat_buf) == -1) {
 459                 if (errno != ENOENT) {
 460                         bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 461                 }
 462                 /* File does not exist */
 463                 return -1;
 464         }
 465         return stat_buf.st_mode;
 466 }
 467
 468 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 469 int unzip_main(int argc, char **argv)
 470 {
 471         enum {
 472                 OPT_l = (1 << 0),
 473                 OPT_x = (1 << 1),
 474                 OPT_j = (1 << 2),
 475         };
 476         unsigned opts;
 477         smallint quiet = 0;
 478         IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 479         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 480         smallint overwrite = O_PROMPT;
 481         uint32_t cdf_offset;
 482         unsigned long total_usize;
 483         unsigned long total_size;
 484         unsigned total_entries;
 485         int dst_fd = -1;
 486         char *src_fn = NULL;
 487         char *dst_fn = NULL;
 488         llist_t *zaccept = NULL;
 489         llist_t *zreject = NULL;
 490         char *base_dir = NULL;
 491         int i;
 492         char key_buf[80]; /* must match size used by my_fgets80 */
 493
 494 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 495  *
 496  * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 497  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 498  * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 499  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 500  * --------  ------  ------- -----   ----   ----   ------    ----
 501  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 502  * --------          -------  ---                            -------
 503  *   204372            35278  83%                            1 file
 504  * # /usr/bin/unzip -v decompress_unlzma.i.zip
 505  * Archive:  decompress_unlzma.i.zip
 506  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 507  * --------  ------  ------- -----   ----   ----   ------    ----
 508  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 509  * --------          -------  ---                            -------
 510  *   204372            35278  83%                            1 file
 511  * # unzip -v decompress_unlzma.i.zip
 512  * Archive:  decompress_unlzma.i.zip
 513  *   Length     Date   Time    Name
 514  *  --------    ----   ----    ----
 515  *    204372  09-06-09 14:23   decompress_unlzma.i
 516  *  --------                   -------
 517  *    204372                   1 files
 518  * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 519  *    204372  09-06-09 14:23   decompress_unlzma.i
 520  * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 521  *   Length     Date   Time    Name
 522  *  --------    ----   ----    ----
 523  *    204372  09-06-09 14:23   decompress_unlzma.i
 524  *  --------                   -------
 525  *    204372                   1 file
 526  * # /usr/bin/unzip -l decompress_unlzma.i.zip
 527  * Archive:  decompress_unlzma.i.zip
 528  *   Length     Date   Time    Name
 529  *  --------    ----   ----    ----
 530  *    204372  09-06-09 14:23   decompress_unlzma.i
 531  *  --------                   -------
 532  *    204372                   1 file
 533  */
 534
 535         opts = 0;
 536         /* '-' makes getopt return 1 for non-options */
 537         while ((i = getopt(argc, argv, "-d:lnopqxjv")) != -1) {
 538                 switch (i) {
 539                 case 'd':  /* Extract to base directory */
 540                         base_dir = optarg;
 541                         break;
 542
 543                 case 'l': /* List */
 544                         opts |= OPT_l;
 545                         break;
 546
 547                 case 'n': /* Never overwrite existing files */
 548                         overwrite = O_NEVER;
 549                         break;
 550
 551                 case 'o': /* Always overwrite existing files */
 552                         overwrite = O_ALWAYS;
 553                         break;
 554
 555                 case 'p': /* Extract files to stdout and fall through to set verbosity */
 556                         dst_fd = STDOUT_FILENO;
 557
 558                 case 'q': /* Be quiet */
 559                         quiet++;
 560                         break;
 561
 562                 case 'v': /* Verbose list */
 563                         IF_FEATURE_UNZIP_CDF(verbose++;)
 564                         opts |= OPT_l;
 565                         break;
 566
 567                 case 'x':
 568                         opts |= OPT_x;
 569                         break;
 570
 571                 case 'j':
 572                         opts |= OPT_j;
 573                         break;
 574
 575                 case 1:
 576                         if (!src_fn) {
 577                                 /* The zip file */
 578                                 /* +5: space for ".zip" and NUL */
 579                                 src_fn = xmalloc(strlen(optarg) + 5);
 580                                 strcpy(src_fn, optarg);
 581                         } else if (!(opts & OPT_x)) {
 582                                 /* Include files */
 583                                 llist_add_to(&zaccept, optarg);
 584                         } else {
 585                                 /* Exclude files */
 586                                 llist_add_to(&zreject, optarg);
 587                         }
 588                         break;
 589
 590                 default:
 591                         bb_show_usage();
 592                 }
 593         }
 594
 595 #ifndef __GLIBC__
 596         /*
 597          * This code is needed for non-GNU getopt
 598          * which doesn't understand "-" in option string.
 599          * The -x option won't work properly in this case:
 600          * "unzip a.zip q -x w e" will be interpreted as
 601          * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 602          */
 603         argv += optind;
 604         if (argv[0]) {
 605                 /* +5: space for ".zip" and NUL */
 606                 src_fn = xmalloc(strlen(argv[0]) + 5);
 607                 strcpy(src_fn, argv[0]);
 608                 while (*++argv)
 609                         llist_add_to(&zaccept, *argv);
 610         }
 611 #endif
 612
 613         if (!src_fn) {
 614                 bb_show_usage();
 615         }
 616
 617         /* Open input file */
 618         if (LONE_DASH(src_fn)) {
 619                 xdup2(STDIN_FILENO, zip_fd);
 620                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 621                 if (overwrite == O_PROMPT)
 622                         overwrite = O_NEVER;
 623         } else {
 624                 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 625                 char *ext = src_fn + strlen(src_fn);
 626                 int src_fd;
 627
 628                 i = 0;
 629                 for (;;) {
 630                         src_fd = open(src_fn, O_RDONLY);
 631                         if (src_fd >= 0)
 632                                 break;
 633                         if (++i > 2) {
 634                                 *ext = '\0';
 635                                 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
 636                         }
 637                         strcpy(ext, extn[i - 1]);
 638                 }
 639                 xmove_fd(src_fd, zip_fd);
 640         }
 641
 642         /* Change dir if necessary */
 643         if (base_dir)
 644                 xchdir(base_dir);
 645
 646         if (quiet <= 1) { /* not -qq */
 647                 if (quiet == 0)
 648                         printf("Archive:  %s\n", src_fn);
 649                 if (opts & OPT_l) {
 650                         puts(verbose ?
 651                                 " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 652                                 "--------  ------  ------- ---- ---------- ----- --------  ----"
 653                                 :
 654                                 "  Length      Date    Time    Name\n"
 655                                 "---------  ---------- -----   ----"
 656                                 );
 657                 }
 658         }
 659
 660 /* Example of an archive with one 0-byte long file named 'z'
 661  * created by Zip 2.31 on Unix:
 662  * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 663  *       sig........ vneed flags compr mtime mdate crc32>
 664  * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 665  *      >..... csize...... usize...... fnlen exlen fn ex>
 666  * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 667  *      >tra_field......................................
 668  * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 669  *       ........... sig........ vmade vneed flags compr
 670  * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 671  *       mtime mdate crc32...... csize...... usize......
 672  * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 673  *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 674  * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 675  *      >..... fn extra_field...........................
 676  * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 677  * 0080  34 00 00 00 00 00                               |4.....|
 678  */
 679         total_usize = 0;
 680         total_size = 0;
 681         total_entries = 0;
 682         cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 683         while (1) {
 684                 zip_header_t zip;
 685                 mode_t dir_mode = 0777;
 686 #if ENABLE_FEATURE_UNZIP_CDF
 687                 mode_t file_mode = 0666;
 688 #endif
 689
 690                 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 691                         /* Normally happens when input is unseekable.
 692                          *
 693                          * Valid ZIP file has Central Directory at the end
 694                          * with central directory file headers (CDFs).
 695                          * After it, there is a Central Directory End structure.
 696                          * CDFs identify what files are in the ZIP and where
 697                          * they are located. This allows ZIP readers to load
 698                          * the list of files without reading the entire ZIP archive.
 699                          * ZIP files may be appended to, only files specified in
 700                          * the CD are valid. Scanning for local file headers is
 701                          * not a correct algorithm.
 702                          *
 703                          * We try to do the above, and resort to "linear" reading
 704                          * of ZIP file only if seek failed or CDE wasn't found.
 705                          */
 706                         uint32_t magic;
 707
 708                         /* Check magic number */
 709                         xread(zip_fd, &magic, 4);
 710                         /* CDF item? Assume there are no more files, exit */
 711                         if (magic == ZIP_CDF_MAGIC) {
 712                                 dbg("got ZIP_CDF_MAGIC");
 713                                 break;
 714                         }
 715                         /* Data descriptor? It was a streaming file, go on */
 716                         if (magic == ZIP_DD_MAGIC) {
 717                                 dbg("got ZIP_DD_MAGIC");
 718                                 /* skip over duplicate crc32, cmpsize and ucmpsize */
 719                                 unzip_skip(3 * 4);
 720                                 continue;
 721                         }
 722                         if (magic != ZIP_FILEHEADER_MAGIC)
 723                                 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 724                         dbg("got ZIP_FILEHEADER_MAGIC");
 725
 726                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 727                         FIX_ENDIANNESS_ZIP(zip);
 728                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 729                                 bb_error_msg_and_die("zip flag %s is not supported",
 730                                         "8 (streaming)");
 731                         }
 732                 }
 733 #if ENABLE_FEATURE_UNZIP_CDF
 734                 else {
 735                         /* cdf_offset is valid (and we know the file is seekable) */
 736                         cdf_header_t cdf;
 737                         cdf_offset = read_next_cdf(cdf_offset, &cdf);
 738                         if (cdf_offset == 0) /* EOF? */
 739                                 break;
 740 # if 1
 741                         xlseek(zip_fd,
 742                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
 743                                 SEEK_SET);
 744                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 745                         FIX_ENDIANNESS_ZIP(zip);
 746                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 747                                 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 748                                  * only from Central Directory.
 749                                  */
 750                                 zip.fmt.crc32    = cdf.fmt.crc32;
 751                                 zip.fmt.cmpsize  = cdf.fmt.cmpsize;
 752                                 zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
 753                         }
 754 // Seen in some zipfiles: central directory 9 byte extra field contains
 755 // a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
 756 // Local header version:
 757 //  u16 0x5455 ("UT")
 758 //  u16 size (1 + 4 * n)
 759 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 760 //  u32 mtime
 761 //  u32 atime
 762 //  u32 ctime
 763 // Central header version:
 764 //  u16 0x5455 ("UT")
 765 //  u16 size (5 (or 1?))
 766 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 767 //  u32 mtime (CDF does not store atime/ctime)
 768 # else
 769                         /* CDF has the same data as local header, no need to read the latter...
 770                          * ...not really. An archive was seen with cdf.extra_len == 6 but
 771                          * zip.extra_len == 0.
 772                          */
 773                         memcpy(&zip.fmt.version,
 774                                 &cdf.fmt.version_needed, ZIP_HEADER_LEN);
 775                         xlseek(zip_fd,
 776                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 777                                 SEEK_SET);
 778 # endif
 779                         if ((cdf.fmt.version_made_by >> 8) == 3) {
 780                                 /* This archive is created on Unix */
 781                                 dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
 782                         }
 783                 }
 784 #endif
 785
 786                 if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
 787                         /* 0x0001 - encrypted */
 788                         bb_error_msg_and_die("zip flag %s is not supported",
 789                                         "1 (encryption)");
 790                 }
 791                 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 792                         (unsigned)zip.fmt.cmpsize,
 793                         (unsigned)zip.fmt.extra_len,
 794                         (unsigned)zip.fmt.ucmpsize
 795                 );
 796
 797                 /* Read filename */
 798                 free(dst_fn);
 799                 die_if_bad_fnamesize(zip.fmt.filename_len);
 800                 dst_fn = xzalloc(zip.fmt.filename_len + 1);
 801                 xread(zip_fd, dst_fn, zip.fmt.filename_len);
 802                 /* Skip extra header bytes */
 803                 unzip_skip(zip.fmt.extra_len);
 804
 805                 /* Guard against "/abspath", "/../" and similar attacks */
 806                 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 807
 808                 if (opts & OPT_j) /* Strip paths? */
 809                         overlapping_strcpy(dst_fn, bb_basename(dst_fn));
 810
 811                 /* Did this strip everything ("DIR/" case)? Then skip */
 812                 if (!dst_fn[0])
 813                         goto skip_cmpsize;
 814
 815                 /* Filter zip entries */
 816                 if (find_list_entry(zreject, dst_fn)
 817                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 818                 ) { /* Skip entry */
 819                         goto skip_cmpsize;
 820                 }
 821
 822                 if (opts & OPT_l) {
 823                         /* List entry */
 824                         char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 825                         sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 826                                 (zip.fmt.moddate >> 5) & 0xf,  // mm: 0x01e0
 827                                 (zip.fmt.moddate)      & 0x1f, // dd: 0x001f
 828                                 (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
 829                                 (zip.fmt.modtime >> 11),       // hh: 0xf800
 830                                 (zip.fmt.modtime >> 5) & 0x3f  // mm: 0x07e0
 831                                 // seconds/2 not shown, encoded in -- 0x001f
 832                         );
 833                         if (!verbose) {
 834                                 //      "  Length      Date    Time    Name\n"
 835                                 //      "---------  ---------- -----   ----"
 836                                 printf(       "%9u  " "%s   "         "%s\n",
 837                                         (unsigned)zip.fmt.ucmpsize,
 838                                         dtbuf,
 839                                         dst_fn);
 840                         } else {
 841                                 char method6[7];
 842                                 unsigned long percents;
 843
 844                                 sprintf(method6, "%6u", zip.fmt.method);
 845                                 if (zip.fmt.method == 0) {
 846                                         strcpy(method6, "Stored");
 847                                 }
 848                                 if (zip.fmt.method == 8) {
 849                                         strcpy(method6, "Defl:N");
 850                                         /* normal, maximum, fast, superfast */
 851                                         IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
 852                                 }
 853                                 percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
 854                                 if ((int32_t)percents < 0)
 855                                         percents = 0; /* happens if ucmpsize < cmpsize */
 856                                 percents = percents * 100;
 857                                 if (zip.fmt.ucmpsize)
 858                                         percents /= zip.fmt.ucmpsize;
 859                                 //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 860                                 //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 861                                 printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 862                                         (unsigned)zip.fmt.ucmpsize,
 863                                         method6,
 864                                         (unsigned)zip.fmt.cmpsize,
 865                                         (unsigned)percents,
 866                                         dtbuf,
 867                                         zip.fmt.crc32,
 868                                         dst_fn);
 869                                 total_size += zip.fmt.cmpsize;
 870                         }
 871                         total_usize += zip.fmt.ucmpsize;
 872                         goto skip_cmpsize;
 873                 }
 874
 875                 if (dst_fd == STDOUT_FILENO) {
 876                         /* Extracting to STDOUT */
 877                         goto do_extract;
 878                 }
 879                 if (last_char_is(dst_fn, '/')) {
 880                         int mode;
 881
 882                         /* Extract directory */
 883                         mode = get_lstat_mode(dst_fn);
 884                         if (mode == -1) { /* ENOENT */
 885                                 if (!quiet) {
 886                                         printf("   creating: %s\n", dst_fn);
 887                                 }
 888                                 unzip_create_leading_dirs(dst_fn);
 889                                 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 890                                         xfunc_die();
 891                                 }
 892                         } else {
 893                                 if (!S_ISDIR(mode)) {
 894                                         bb_error_msg_and_die("'%s' exists but is not a %s",
 895                                                 dst_fn, "directory");
 896                                 }
 897                         }
 898                         goto skip_cmpsize;
 899                 }
 900  check_file:
 901                 /* Does target file already exist? */
 902                 {
 903                         int mode = get_lstat_mode(dst_fn);
 904                         if (mode == -1) {
 905                                 /* ENOENT: does not exist */
 906                                 goto do_open_and_extract;
 907                         }
 908                         if (overwrite == O_NEVER) {
 909                                 goto skip_cmpsize;
 910                         }
 911                         if (!S_ISREG(mode)) {
 912  fishy:
 913                                 bb_error_msg_and_die("'%s' exists but is not a %s",
 914                                         dst_fn, "regular file");
 915                         }
 916                         if (overwrite == O_ALWAYS) {
 917                                 goto do_open_and_extract;
 918                         }
 919                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 920                         my_fgets80(key_buf);
 921                         /* User input could take a long time. Is it still a regular file? */
 922                         mode = get_lstat_mode(dst_fn);
 923                         if (!S_ISREG(mode))
 924                                 goto fishy;
 925                 }
 926
 927                 /* Extract (or skip) it */
 928                 switch (key_buf[0]) {
 929                 case 'A':
 930                         overwrite = O_ALWAYS;
 931                 case 'y': /* Open file and fall into unzip */
 932  do_open_and_extract:
 933                         unzip_create_leading_dirs(dst_fn);
 934 #if ENABLE_FEATURE_UNZIP_CDF
 935                         dst_fd = -1;
 936                         if (!S_ISLNK(file_mode)) {
 937                                 dst_fd = xopen3(dst_fn,
 938                                         O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW,
 939                                         file_mode);
 940                         }
 941 #else
 942                         /* O_NOFOLLOW defends against symlink attacks */
 943                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW);
 944 #endif
 945  do_extract:
 946                         if (!quiet) {
 947                                 printf(/* zip.fmt.method == 0
 948                                         ? " extracting: %s\n"
 949                                         : */ "  inflating: %s\n", dst_fn);
 950                         }
 951 #if ENABLE_FEATURE_UNZIP_CDF
 952                         if (S_ISLNK(file_mode)) {
 953                                 if (dst_fd != STDOUT_FILENO) /* not -p? */
 954                                         unzip_extract_symlink(&zip, dst_fn);
 955                         } else
 956 #endif
 957                         {
 958                                 unzip_extract(&zip, dst_fd);
 959                                 if (dst_fd != STDOUT_FILENO) {
 960                                         /* closing STDOUT is potentially bad for future business */
 961                                         close(dst_fd);
 962                                 }
 963                         }
 964                         break;
 965
 966                 case 'N':
 967                         overwrite = O_NEVER;
 968                 case 'n': /* Skip entry data */
 969  skip_cmpsize:
 970                         unzip_skip(zip.fmt.cmpsize);
 971                         break;
 972
 973                 case 'r':
 974                         /* Prompt for new name */
 975                         printf("new name: ");
 976                         my_fgets80(key_buf);
 977                         free(dst_fn);
 978                         dst_fn = xstrdup(key_buf);
 979                         chomp(dst_fn);
 980                         goto check_file;
 981
 982                 default:
 983                         printf("error: invalid response [%c]\n", (char)key_buf[0]);
 984                         goto check_file;
 985                 }
 986
 987                 total_entries++;
 988         }
 989
 990         if ((opts & OPT_l) && quiet <= 1) {
 991                 if (!verbose) {
 992                         //      "  Length      Date    Time    Name\n"
 993                         //      "---------  ---------- -----   ----"
 994                         printf( " --------%21s"               "-------\n"
 995                                      "%9lu%21s"               "%u files\n",
 996                                 "",
 997                                 total_usize, "", total_entries);
 998                 } else {
 999                         unsigned long percents = total_usize - total_size;
1000                         if ((long)percents < 0)
1001                                 percents = 0; /* happens if usize < size */
1002                         percents = percents * 100;
1003                         if (total_usize)
1004                                 percents /= total_usize;
1005                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
1006                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
1007                         printf( "--------          ------- ----%28s"                      "----\n"
1008                                 "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
1009                                 "",
1010                                 total_usize, total_size, (unsigned)percents, "",
1011                                 total_entries);
1012                 }
1013         }
1014
1015         return 0;
1016 }